Do we know how badly we can stare at dots? Evidence from double step saccades and a universal analysis method.

Staring at dots seems easy enough. Is it really? The visual system likes to represent spatial locations of objects retinotopically, that is, relative to the center of gaze. This means that every gaze shift (eye movement) requires all object coordinates to be updated to reflect their new retinotopic location. Without this process (spatial remapping) we would lose our sense of where things are every time we move our eyes (make a saccade).
In this paper we asked the question whether remapping occurs based on the visual vector (the desired eye displacement) or the actual motor vector (or a copy thereof: corollary discharge), which typically doesn't fully match the visual vector. Previous research investigated this using the double step saccade (DSS) task. Two dots are flashed briefly and participants make two eye movements to the remembered locations of these dots. To hit the first is easy, almost reflexive. The second has to be targeted from memory and its location coordinates updated after the first eye movement. Note that the screen is black at this point and provides no visual guidance, the visual system must rely on its internal computations to produce an accurate second saccade vector. Here we expand the DSS task to saccades in all directions and random amplitudes. We produced an analysis method that allows for this more realistic task design. Endpoint errors in the first saccade create a mismatch between the desired and actual first saccade vector. The relation between these endpoint errors and the direction of the second saccade vector provides a direct percent estimate of the degree to which the visual system is aware of and corrects for endpoint errors in the first saccade vector. We find around 70% correction for first saccade endpoint errors in second saccade vectors.  
We further applied this paradigm to investigate how saccade adaptation (a global change to saccade amplitudes) influences error correction. Our data demonstrates that endpoint errors caused by induced adaptation are accounted for in the remapping process. In other words, corollary discharge (CD) vectors contain adapted vectors. This implicates that at least some CD vectors are taken from sites downstream of adaptation (i.e. downstream from the cerebellum).
We developed a more generalizable DSS paradigm and analysis method, verify earlier research with more restricted DSS tasks and apply the paradigm to the question whether corollary discharge contains adapted vectors.